Noise, resolution and sensitivity
These images show 72ppi (100% on a computer screen) sections of images of a resolution chart, captured with the camera’s lens set to around 105mm. We show the section of the resolution chart where the camera starts to fail to reproduce the lines separately. The higher the number visible in these images, the better the camera’s detail resolution is at the specified sensitivity setting.
Our resolution chart test shows that in good light, the Canon PowerShot G15 is capable of resolving about as much detail as a 12-million-pixel DSLR, even when shooting JPEG files. At ISO 80-200, the camera can resolve up to 24 on our chart, and the fall-off in detail as the sensitivity increases is fairly slight. It is only when ISO 3200 is reached that there is a noticeable loss in the quality of JPEG images. At this point the luminance and colour noise reduction starts to have a real effect in textured areas, causing softening and loss of detail.
At the highest ISO 12,800 sensitivity, colour and luminance noise, as well as areas of noise reduction, are clearly visible. As is often the case, this maximum setting should only really be used as a last resort. However, we should still remember that the fact a compact camera is capable of shooting at such a high sensitivity setting is impressive in itself.
I found the range between ISO 80 and ISO 400 gave very clean results. There is noise in shadow areas, which becomes more prominent if the images are adjusted in editing software, but on the whole they are crisp.
It is at ISO 800 when shadow noise really begins to creep in. Looking back on images taken with the G12, I would say there is a slight improvement, with shots taken at ISO 800 on the G15 looking more like those at ISO 400 on the G12.
As would be expected, more detail can be extracted from raw files, and the rather basic preset settings of the Canon DPP software can actually make significant improvements, particularly to sharpness. Colour noise can also be tackled more easily using the software.
